Gallops research lab
What the lab wants to know next. An investigator raises a question when an observation surprises it, and ranks it by how much an answer would teach. Each links to the thread it belongs to and the observations that raised it; the highest priority comes first.
Loading questions…
Does the GB-only version of the live H199 rule hold up over the full history from 2020, or shrink towards par as the pooled second-favourite rival did (1.158 to 1.04)?
open0.80This decides whether SYS-big-price-favourite-big-handicap should be judged and staked on its GB bets only. Columns: course, is_favourite, sp_decimal, field_size, is_handicap, book_pct.
Thread: H199: favourite at a big price in a big handicaprace-bettingraised 16 Oct 2026
Does the live H199 rule (favourite at 5/1+ in a 14+ runner handicap) also fall towards 1.04 over the full history from 2020, as its second-favourite rival did in the vault?
open0.80Second favourites went from 1.158 in discovery to 1.04 on 4,207 bets in the vault. If H199 shrinks the same way, the live system's expected edge is near zero after commission, and its forward record should be judged against about 1.04, not 1.13. Columns: is_handicap, is_favourite, sp_decimal, field_size, read across 2020-07 onwards with the period split.
Thread: H199: favourite at a big price in a big handicaprace-bettingraised 16 Oct 2026
Why does the lab's discovery measurement (A/E 0.963 on 204) disagree with the imported GREEN (1.095 on 435)? Is it the field-size-corrected yardstick, the window, or the edge fading?
testing0.80If the field-size-corrected yardstick explains the gap, the live system rests on a mismeasurement and should be reviewed. Compare the same rule under both yardsticks and against the other runners in the same races, splitting by field_size and sp_rank. Read field_size, sp_decimal, sp_rank and region.
Thread: H45: young horses in novice chasestested by Young horses in handicap chases also beat their priceageraised 16 Oct 2026
Is the period fade in big-handicap top-two pools (1.33 Oct–Apr, 0.92 after) really a code effect: lifted in jumps handicaps and at par in flat handicaps?
testing0.75The two periods line up with the jumps winter and the flat summer. If code is the lever, the live H199 system and every rule in this thread should be restricted to jumps, and the vault shrinkage is explained as season-mixing rather than decay. A test reads race_type, code, is_handicap, sp_rank, sp_decimal and field_size, and splits by race_type within each period.
Thread: H199: favourite at a big price in a big handicaptested by Rival: the big-handicap front-of-market lift is a jumps effect, and the period fade is the switch to the flatrace-bettingraised 16 Oct 2026
Is the big-yard top-two lift in big handicaps carried by a few trainers (for example two Irish mega-yards), or spread across many big yards in both countries?
testing0.75An effect resting on one or two trainers is a yard-specific pricing quirk, not a general big-yard mechanism, and it is fragile going forward. A per-trainer contribution table over the H-659f3a37 pool settles it. Columns: trainer, trainer_id, trainer_runs, sp_rank, sp_decimal, field_size, is_handicap, course.
Thread: H199: favourite at a big price in a big handicaptested by Rival: yard quality, not yard size, drives the big-handicap top-two liftrace-bettingraised 16 Oct 2026
Is the live H199 system's lift also carried by GB alone, with Irish big-handicap favourites at 5/1+ sitting at par, as the second-favourite rival showed (GB 1.40 vs IRE 0.99)?
testing0.75If yes, the live system should be judged on its GB bets only, and the mechanism is about how GB big-handicap markets form rather than about favourite status. Needs H199's by_country breakdown, or a course_id-based split, because there is no country column. Columns: is_favourite, sp_decimal, field_size, is_handicap, course_id. This is the cheapest high-value check because it re-reads an existing result.
Thread: H199: favourite at a big price in a big handicaptested by Replicate: H199 favourites at 5/1+ in 14+ runner handicaps at GB courses onlyrace-bettingraised 16 Oct 2026
Why did H-da5bbd88's vault query return 0 bets? Is last_non_finish_code ('every non-finisher is PU before 2025') structurally empty in the vault window, which would make every BD/UR vault verdict void?
open0.75A 'supportive' vault verdict on 0 bets moves a hypothesis along without evidence. If the vault window falls before 2025-01-01, every rule on last_non_finish_code, last_fell_or_unseated or last_fell_going_well (from 2025-11) will be vacuous. A check should count non-null last_non_finish_code values by month inside the vault window, alongside last_code and days_since_last_run. That answer decides whether this family's vault stage means anything.
Thread: H98: brought down or unseatedlettersraised 16 Oct 2026
Do all runners in non-handicap novice chases beat the field-size-corrected yardstick, so that H45's imported lift is a property of the race type rather than of age?
testing0.75The handicap generalisation failed: 0.967 on 521, with a same-race control of 1.017 (O-4dc10e5b). So youth does not travel outside novice chases. If older runners in the same novice chases also sit near 1.09, the live system SYS-young-novice-chasers is selecting a race type, not a horse, and needs review. Read race_type, is_novice, is_handicap, age, field_size, sp_rank and book_pct, and split by region and period.
Thread: H45: young horses in novice chasestested by Older horses in non-handicap novice chases match the young ones (race-type rival to H45)ageraised 16 Oct 2026
Why did the GB-only H199 replicate return only 48 discovery bets? Is the Irish-course exclusion list in `course` removing GB races by mistake (a text or case mismatch)?
open0.70H199 had 556 discovery bets, and GB made up about 30% of every split, so a GB-only version should give about 150 or more, not 48. Until this is fixed, Q-701eba5c and Q-c7c25090 cannot be answered. Columns: course, course_id.
Thread: H199: favourite at a big price in a big handicaprace-bettingraised 16 Oct 2026
Is the Snowden novice non-handicap lift spread across the price range, or carried by one short-price band the way the marathon slices were?
testing0.70Other families showed single-band concentration (4/1 to 7/1) that killed apparent edges. A trainer edge from undervalued development should appear at longer prices too, where a horse's improvement is least priced in. Columns: sp_decimal, sp_rank, trainer, is_novice, is_handicap, race_type.
Thread: H17: jamie snowden novicestested by Rival: Snowden novice lift is a short-price artefact; test the 9/2+ runners on their ownnamed-trainerraised 16 Oct 2026
Does the live H199 rule show the same early-period concentration and after-2026-04-27 fade as the big-yard replicate?
open0.65If H199's forward bets since April score at par, the live system is running on a decayed or seasonal edge. A test reads is_favourite, sp_decimal, field_size and is_handicap, split by date and race_type.
Thread: H199: favourite at a big price in a big handicaprace-bettingraised 16 Oct 2026
Is the broad H126 lift (1.13 on 1904 at discovery) decaying over time, given that the March–April sharpen fell to 1.02 on 1548 in the vault?
open0.65The vault shrinkage could be regression from a lucky discovery slice, or a real fade as the market learns. A season-by-season split of the base rule would tell us which and decide whether to keep investing in the thread. Columns: race_type, is_handicap, month, age, prior_runs, days_since_last_run, plus the race-date year.
Thread: H126: lightly raced four year old returnsseasonalraised 16 Oct 2026
Is the price-capped stiff-sprint edge (1.64 on 141 in the vault) spread across the stiff tracks, or carried by one or two courses or a draw bias at those courses?
testing0.65Only a handful of GB tracks have a stiff finish and a sprint course. A 141-bet edge could be one course's quirk, such as a draw bias on a 5f straight, rather than the hill-adds-stamina mechanism. If one course carries it, the live system is a course angle, not a mechanism. Columns: course, course_id, draw, track_draw_bias, track_straight_furlongs, field_size.
Thread: H181: proven over further at a stiff tracktested by Rival: the stiff-sprint edge is a course or draw quirk, not the hilltrack-staminaraised 16 Oct 2026
Are small-yard Irish runners in the top two at 5/1+ in big handicaps systematically over-bet (A/E around 0.88), and is that a stable negative effect rather than noise?
open0.60This is the other half of the IRE split (0.88 on 328, against 1.38 for big Irish yards). If it is stable, it explains every earlier finding of Ireland at par, and it is a lay or avoid signal. It needs a reliable IRE flag; the course-list approach returned only 48 bets (see Q-c1364827). Columns: course, course_id, trainer_runs, sp_rank, sp_decimal, field_size, is_handicap.
Thread: H199: favourite at a big price in a big handicaprace-bettingraised 16 Oct 2026
Is the lift strongest on a novice's first one or two runs over obstacles, where the form book holds the least about the horse?
testing0.60If the mechanism is that the yard's horses improve more than visible form suggests, the market should be most wrong when there is little hurdle or chase form. Columns: hurdle_runs, chase_runs, prior_runs, race_type, days_since_last_run.
Thread: H17: jamie snowden novicestested by Sharpen: Snowden novices with at most one prior run in the codenamed-trainerraised 16 Oct 2026
Is the H19 lift a turf-season effect? Handicap H19 scored 1.29 after late April 2026 but 0.96 from October to April, when Flat racing is mostly all-weather.
testing0.60The handicap sharpen's period split (1.29 against 0.96 on 78) lines up with the all-weather winter and the turf summer. If checks on all-weather tracks are common, and race-readers and the market allow for them, the trouble edge would only exist on turf. A turf-only result would explain the uneven periods; a split by surface with no difference would rule this out. Columns: surface, last_surface, race_type, last_momentum_interrupted, last_beaten_lengths, last_position.
Thread: H19: momentum interrupted beaten narrowlytested by Rival: the H19 lift lives on turf, not all-weatherstoryraised 16 Oct 2026
Do big-field festival marathon handicaps (Nationals and similar) produce the lift, rather than youth?
testing0.60Marathon handicap chases are dominated by a few large, competitive, high-overround races, where longshot pricing behaves differently. Columns: field_size, book_pct, course, race_class, sp_decimal. If the lift disappears in ordinary-sized fields, the system is a race-type effect.
Thread: H175: young horse in a marathon chasetested by Rival: big-field festival marathon handicaps, not youth, produce the lifttested by Young marathon handicap chasers keep the lift outside big-field racesage-staminaraised 16 Oct 2026
In 3m4f+ handicap chases, are runners within 7 lb of bottom weight (all ages) over-bet, and is that the true source of the weight-position gap?
testing0.60Same-race light-weights scored 0.60 against 1.21 for higher weights. Older light-weights scored 0.48 on 124 and young ones 0.80 on 37. Together these give roughly 160+ bets on the negative side. A test would read weight_lbs, bottom_weight_lbs, weight_rank_low, sp_decimal and field_size.
Thread: H175: young horse in a marathon chasetested by Near-bottom-weight runners of all ages in marathon handicap chases are over-betage-staminaraised 16 Oct 2026
Within turf-to-AW Flat handicap switchers, does A/E rise steadily with the length of the layoff (14-60, 61-149, 150+ days), as a turf-discount-plus-readiness mechanism predicts?
testing0.60Short gap scored 1.07 on 4045, 150+ days 1.23 on 654 and 120+ days 1.295 on 899. A smooth gradient would support a two-part cause. A flat middle would mean the 150+ result is a lucky slice. Columns: days_since_last_run, surface, last_surface, is_handicap, region.
Thread: H126: lightly raced four year old returnstested by Dose-response: turf-to-AW handicap switchers after a 61-149 day gap sit between the short- and long-gap liftsseasonalraised 16 Oct 2026
Is the drift lift specific to the wide tag, or do all close-beaten runners whose price drifted since last time beat their SP?
testing0.60If close-beaten drifters without the wide tag also score well above 1.0, the edge is a general overreaction to drift and the wide tag is incidental. Read last_raced_wide, last_position, last_beaten_lengths, sp_decimal and last_sp_decimal. The shortened complement at 1.165 against drifters at 1.47 shows that drift matters inside the wide base, but not whether the wide tag does.
Thread: IQ140a: wide trip with figurestested by Rival: close-beaten drifters pay without the wide tagoverlayraised 16 Oct 2026
Do lightly raced four-year-old returners whose mark was raised during their three-year-old season carry the H126 edge, as an improver-trajectory mechanism would predict?
testing0.60The unraised-mark cut came out the opposite way (0.95 on 151), while the won-last-time system scores 1.27. If raised horses carry the lift, the mechanism is the market doubting continued progression, not a stale mark. A test reads rating_change, last_official_rating, last_position, prior_runs, days_since_last_run and age.
Thread: H126: lightly raced four year old returnstested by Lightly raced four-year-old returners whose mark was raised: improvers the market doubtsseasonalraised 16 Oct 2026
Is the young-novice-chaser lift concentrated on chase debuts (chase_runs = 0), where only the yard knows how well the horse jumps?
testing0.60This is a direct test of the mechanism. The market can see a young horse's later chase runs, but not its schooling before its first. Read chase_runs, chase_wins, age and jumping_errors_last3. It should be set against H124, where fit hurdlers on their chase debut were over-bet (0.91).
Thread: H45: young horses in novice chasestested by Young novice chasers on their chase debut beat their priceageraised 16 Oct 2026
Is the short-end lift in big handicaps driven by the bookmakers' overround, so that it is strongest where book_pct is highest and the SP to BSP gap widest?
open0.55Both the favourite and second-favourite effects appear at BSP in big fields, where SP overround is largest. If the effect tracks book_pct, it is a stretched-SP-book artefact that pays only on the exchange. If it does not, something about the horses is under-priced. Columns: book_pct, field_size, sp_rank, sp_decimal, favourite_sp, is_handicap.
Thread: H199: favourite at a big price in a big handicaprace-bettingraised 16 Oct 2026
When the troubled run came on turf, does the H19 lift hold in both periods, while checks on all-weather tracks sit at par?
open0.55The handicap sharpen died on a period split: 0.96 from October to April against 1.29 after late April, the line between the all-weather winter and the turf summer. Being checked is routine on tight all-weather bends, so race-readers and the market may already allow for it there; on turf it is rarer and more telling. Columns: last_surface, surface, last_momentum_interrupted, last_beaten_lengths, last_position, is_handicap.
Thread: H19: momentum interrupted beaten narrowlystoryraised 16 Oct 2026
Is the young marathon-chase lift specific to horses still unexposed over fences, whose chase form understates how much they will improve at an extreme trip?
testing0.55The hurdle version failed (0.945 on 772), so the edge seems to need fences. If it comes from unexposed chasers improving, it should be stronger with few chase runs and weaker in exposed young chasers. Columns: chase_runs, chase_wins, age, distance_yards, is_handicap, prior_runs.
Thread: H175: young horse in a marathon chasetested by Young marathon handicap chasers with little chase experience carry the lifttested by Exposed young marathon handicap chasers sit at par: the lift needs an unexposed chaserage-staminaraised 16 Oct 2026
Is the gap between discovery and vault in this thread a time effect (the market has priced turf-to-AW switchers since about 2024) or regression from selecting the best slice?
open0.55Two vault results (1.09→1.03 and 1.14→0.98) shrank sharply in the same direction. Splitting the turf-to-AW 150+ rule season by season (columns: last_surface, surface, days_since_last_run, is_handicap, race_type, region) would show whether discovery-era years are high and later years flat (a decay) or whether the lift is scattered noise. The answer decides whether to retire the AW branch for good.
Thread: H126: lightly raced four year old returnsseasonalraised 16 Oct 2026
Within spring Flat handicap returners after 150+ days, does the lift sit in the second tier of the market (roughly 3/1 to 7/1) rather than at 3/1 or shorter, at any age?
testing0.55The sp_decimal <= 5 neighbour of H-71a94a07 scored 1.214 on 273 with periods 1.22 and 1.20. Its 4/1-to-7/1 band scored 1.42 on 132. The up-to-3/1 rule scored 1.10, and the evens-or-shorter band 0.71 on 11. H-1a6f83c9 (14/1 or shorter) also scored 1.21. If the lift lives in the mid-price band, the thread's 'short band carries it' reading was mislocated. A test reads sp_decimal, sp_rank, days_since_last_run, is_handicap, month and age.
Thread: H126: lightly raced four year old returnstested by Second-tier fancied spring handicap returners (3/1 to 7/1) beat their prices at any ageseasonalraised 16 Oct 2026
Why do Irish stiff-track sprints show no lift (IRE 0.85 on 23) while GB shows 1.75 on 60? Is track_stiff_finish coding a different kind of finish in Ireland, or are the GB bets concentrated on one or two courses?
testing0.55If the IRE stiff flag marks courses with only a mild rise, the mechanism predicts no lift there, and the live system should perhaps be GB only. If GB is one course, Q-f268fe31's course-quirk worry stands. A per-course descriptive scan should come before any rule. Columns: region, course, track_stiff_finish, track_finish, track_undulations, track_straight_furlongs, distance_yards, last_distance_yards, sp_decimal.
Thread: H181: proven over further at a stiff tracktested by Rival: the stiff-sprint lift is one or two GB courses, not the hilltrack-staminaraised 16 Oct 2026
Is the lift after April 2026, seen in several wide-trip subsets, a summer turf effect, or a change in how often and where the 'wide' tag gets recorded?
testing0.55O-b4f84fea shows the small-field non-AW subset at 0.92 before 2026-04-27 and 1.47 after. Earlier findings note the same lift elsewhere. If the tag rate changed (read last_raced_wide share by month, last_surface, race_type, region), the base edge may not be stable going forward. A jumps-today check (race_type) separates season from tagging.
Thread: IQ140a: wide trip with figurestested by Wide-trip base in jump races today: is the edge only a summer Flat turf effect?overlayraised 16 Oct 2026
Among quick-returning last-time winners at 9/1+, is the shortfall concentrated in horses whose previous win came at a long price, a surprise win the market now over-trusts?
testing0.50The thread has ruled out the mark rise and the quick return as separate causes, which leaves over-reaction to a fresh win. If so, the deficit should be largest where the win was least expected and most likely flattered by circumstance. A descriptive scan should split the H-9e92ea71 pool by last_sp_decimal (e.g. up to 4, 4 to 9, over 9), using last_position, days_since_last_run and sp_decimal, and could also check l_time_index or l_fsp_pct for how strong the winning run really was.
Thread: Raised handicap winners sent off at 9/1+ are over-bet, at any gap since the wintested by Test-mechanism: quick-returning winners at 9/1+ whose last win came at 7/1+ are over-betnegativeraised 16 Oct 2026
Does the market's over-fade of big handicap rises persist when the horse returns 15 to 35 days after the win, or is it tied to a quick turnout?
testing0.50Big rises lift both easy winners (1.47 on 131) and non-easy winners (1.13 on 318) back within 14 days. If the cause is punters anchoring on the pounds raised, the lift should hold at any interval. If it is about catching a horse at its peak, it should vanish after 14 days, as the easy-winner 15 to 35 day slice did (1.013 on 636). This decides whether the system is a 'big rise' rule or a 'quick turnout' rule. Columns: rating_change, days_since_last_run, last_position, is_handicap, last_is_handicap, last_won_easily.
Thread: H34: won easily last timetested by Rival: big-rise handicap winners beat their price at 15 to 35 days too, if the market over-fades the risestoryraised 16 Oct 2026
Does the cumulative trainer_runs >= 1500 filter bias the big-yard pool towards later dates and the largest yards? Does a fixed-size definition, such as trainer_runs >= 3000, give the same answer?
testing0.50trainer_runs grows from the 2020 data start, so the set of yards that qualify changes over time. In the vault, back to 2020, only a few yards will qualify early on, so the vault test would be measuring a different population. Columns: trainer_runs, trainer_kind_runs, dv_yard_n, trainer_id.
Thread: H199: favourite at a big price in a big handicaptested by Replicate big-yard top-two lift with a fixed, time-stable definition of big yardrace-bettingraised 16 Oct 2026
Why would GB big-handicap top-of-market runners beat their prices while Irish ones do not? Is the Irish market better informed (big-yard money), or do Irish big fields carry a lower overround per runner?
open0.50A mechanism for the country split decides whether the GB lift is real structure or noise. Compare book_pct, favourite_sp and field_size between GB and Irish 14+ runner handicaps. Also compare trainer concentration (trainer_id share among the top two) as a sign of how informed the market is.
Thread: H199: favourite at a big price in a big handicaprace-bettingraised 16 Oct 2026
Should child tests in this thread be judged on the full 2020-on history (where the parent had 282 to 488 bets) instead of the October 2025 discovery window, since no sub-slice can reach 150 bets there?
open0.50Three children (16, 27 and 35 bets) all pointed positive and were all killed or made niche by sample size alone. In this window every slice of a single-trainer system is untestable, and the verdicts carry no information about the mechanism. Columns: trainer, is_novice, hurdle_runs, chase_runs, sp_decimal.
Thread: H17: jamie snowden novicesnamed-trainerraised 16 Oct 2026
Does the momentum-interrupted, narrowly-beaten lift hold in hurdles and chases, or only on the Flat where discovery was strongest?
testing0.50On the Flat, races are short and tactical, so a check mid-run is decisive. Over jumps, in longer races, horses have more time to recover. Columns: race_type, last_race_type, last_momentum_interrupted, last_beaten_lengths.
Thread: H19: momentum interrupted beaten narrowlytested by Replicate: momentum interrupted, beaten 2 lengths or less, in hurdles and chasesstoryraised 16 Oct 2026
Are the up-to-7/1 winners that carry every killed marathon slice the same small set of races and horses, so the four 'concentration' kills are one observation, not four?
open0.50Four slices each failed on the same band (1.33 to 1.68 up to 7/1, rest 0.89 to 0.95). If they share most winners, the family has one short-price cluster and the parent's support is narrower than it looks. A descriptive overlap check would read sp_decimal, sp_rank, horse_id, course, field_size and age within race_type='Chase', is_handicap=1, distance_yards>=6160.
Thread: H175: young horse in a marathon chaseage-staminaraised 16 Oct 2026
Is the turf-to-AW lift really a class drop, with turf horses dropping into weaker AW handicaps, rather than a surface discount?
testing0.50Turf-to-AW switchers often drop in class, and that drop is visible but may be under-weighted. If switchers who do not drop in class score about 1.0, the surface story fails. GB only, because IRE race_class only starts in 2026-07. Columns: race_class, last_race_class, surface, last_surface, region.
Thread: H126: lightly raced four year old returnstested by Rival: turf-to-AW switchers that are not dropping in class are fairly priced, so a class drop drives the liftseasonalraised 16 Oct 2026
Is the weak-visible-form lift (best_last3_beaten > 5 at 14/1 or shorter) specific to horses that departed last time, or shared by any jumps runner with poor recent form at a short price?
testing0.50H-e7675af7's niche rests on a flag that adds nothing. A general 'poor form but fancied' effect (the market knows something, e.g. a drop in class or a yard's intent) would produce the same number. Columns: best_last3_beaten, sp_decimal, race_type, last_fell_or_unseated.
Thread: H98: brought down or unseatedtested by Rival: poor recent form at 14/1 or shorter beats its price without any departurelettersraised 16 Oct 2026
Does the spring returner lift need a handicap (a mark that lags the horse's progress), or does a rust discount alone produce it in non-handicaps too?
testing0.50Dropping is_handicap from H-71a94a07 cut A/E from 1.10 on 141 to 1.02 on 421. That suggests non-handicap fancied returners are priced about right. If confirmed, the mechanism is about the handicap mark, not market rust, and the thread should look at last_official_rating and days_since_win rather than price. A test reads is_handicap, is_maiden, is_novice, days_since_last_run and sp_decimal.
Thread: H126: lightly raced four year old returnstested by Fancied spring returners in non-handicaps are priced right: the lift needs a handicap markseasonalraised 16 Oct 2026
Does the stiff-sprint edge grow with the size of the trip drop (a 660+ yard drop beating a 330 to 659 yard drop), as the stamina-becomes-speed mechanism needs?
testing0.50The vault figure (1.17 on 415) is weak enough that we need a different consequence of the same cause, not another slice. If stamina is the cause, the size of the drop should rank the A/E. Columns: last_distance_yards, last_win_yards, distance_yards, track_stiff_finish, sp_decimal, race_type. A flat or reversed ranking would point to general trip-drop drift.
Thread: H181: proven over further at a stiff tracktested by Dose-response: a big trip drop (660+ yards) into a stiff Flat sprint pays moretrack-staminaraised 16 Oct 2026
Does a quick return after an unseat or brought-down beat a slow return, as the 'had no race, still fresh' mechanism requires?
testing0.50The mechanism is that the horse kept its fitness and form because it did no real racing. That predicts the lift is concentrated where `days_since_last_run` is short (28 days or less). After a long break, the incident no longer matters. Columns: `last_non_finish_code`, `days_since_last_run`, `race_type`. H118 (early exit within 28 days, chase) was killed at 1.049 on 76, so this needs the broader population.
Thread: H98: brought down or unseatedtested by Unseated or brought down and back within four weeks: the fresh-horse consequencelettersraised 16 Oct 2026
Do lightly raced three-year-old handicap returners in spring show the same lift, or is the mispricing specific to four-year-olds?
open0.50The older-horse control (O-6ea233f1, 0.64 on 119) points toward an age-specific effect but is underpowered. Three-year-olds are the stronger contrast: they mature over the winter even more, yet the market expects them to improve. If they score about 1.0 while four-year-olds score 1.13, the edge lies in what the market expects of four-year-olds rather than in maturation alone. A test reads age, prior_runs, days_since_last_run, month and is_handicap.
Thread: H126: lightly raced four year old returnsseasonalraised 16 Oct 2026
Within the H126 handicap base, do returners priced 16/1 or bigger (sp_decimal > 15) score at or below 0.90, or as well as the capped group?
testing0.50The capped rule's planned complement control was never run. Unfancied returners across both race kinds scored 1.11 on 450, while earlier side cuts blanked at 16/1+. A test reads sp_decimal, sp_rank and is_handicap within the H126 base. The answer decides whether a price filter belongs in the published system at all.
Thread: H126: lightly raced four year old returnstested by Long-priced four-year-old handicap returners: is the 16/1+ tail priced right?seasonalraised 16 Oct 2026
Is the drift edge carried by a change of going since last time, i.e. does the market over-penalise a switch of ground for close-beaten horses?
testing0.50The same-going refinement of H-c869bcd0 fell to 1.245 on 120 (p 0.21, periods 0.96/1.69), so the roughly 140 going-changed drifters must score near 1.65. Read going_group, last_going_group, sp_decimal, last_sp_decimal and last_beaten_lengths. This decides whether the mechanism is reading the wide run or reading the ground.
Thread: IQ140a: wide trip with figurestested by Rival: drift after a going change is over-penalised for close-beaten runnersoverlayraised 16 Oct 2026
Does the small-field-last into bigger-field-today lift need the wide tag, or do all close-beaten runners stepping up in field size beat their price?
testing0.50H-dc80c356 (O-0d2a2141) scored 1.57 but holds two conditions at once. Moving from a small field to a big one may be underbet in general, because small-field form looks tactical. Read last_field_size, field_size, last_position, last_beaten_lengths, with and without last_raced_wide.
Thread: IQ140a: wide trip with figurestested by Rival: close-beaten runners stepping up from small to big fields, wide tag removedoverlayraised 16 Oct 2026
Does the four-year-old returner lift depend on being lightly raced, or do exposed four-year-olds (more than 10 prior runs) back after 150+ days in spring handicaps score the same?
testing0.50Both surfaces met the vault at about 1.10 (O-3b4db8ee, O-bbb9c77c), and the stale-mark story failed. If exposed four-year-olds also score about 1.10, the cause is a spring-freshness mispricing for this age, not unexposed improvement. Columns: age, prior_runs, days_since_last_run, is_handicap, month, race_type.
Thread: H126: lightly raced four year old returnstested by Exposed four-year-olds back from the winter in spring handicaps: is light racing the cause?seasonalraised 16 Oct 2026
If last-time track sharpness makes no difference, does 'raced wide' still pay where there are no bends, i.e. after turf sprints on straight courses?
testing0.50The ground-on-bends mechanism predicted galloping < sharp. It came out 1.23 against 1.18, the wrong way. If wide runners in straight-course sprints also beat their price, the tag is marking something else: racing without cover, or being on the wrong part of the track. In that case the mechanism text of the published system is wrong even though the edge is real. Columns: last_raced_wide, last_distance_yards, last_race_type, last_surface, last_track_straight_furlongs, last_track_sharpness.
Thread: IQ140a: wide trip with figurestested by Wide trip after a turf sprint (no bends) still pays: the tag is not about lost groundtested by Wide and prominent last time: covered extra ground while doing the pace work
Are last-time winners whose win came from an uncontested, soft lead over-bet next time, when the speed figures alone are not?
testing0.45A slow top speed is only an indirect sign of a tactical gift. l_uncontested, l_pressed and last_led_in_running measure it directly. A scan could compare A/E for last_position = 1 by l_uncontested and l_pressed, with sp_rank <= 3, and also check l_time_vs_par.
Thread: Last-time winners that were slow in the race are over-bet at the front of the markettested by Winners gifted an uncontested lead are over-bet at the front of the market next timenegativeraised 16 Oct 2026
Is the 9/1–14/1 deficit (0.78 on 966) a flat or a jumps effect, given that the later, flat-heavy period scored 0.79 and the earlier, winter-jumps period 0.93?
open0.45The two periods differ in code mix, so the deficit may belong to one code only. A descriptive scan by race_type and region on the raised-winner pool would show this before any rule is written. Columns: race_type, region, sp_decimal, rating_change, days_since_last_run.
Thread: Raised handicap winners sent off at 9/1+ are over-bet, at any gap since the winnegativeraised 16 Oct 2026
Is the market's over-fade about the pounds raised, so it also shows on horses raised 6 lb or more after finishing second, without a win?
testing0.45This separates two readings. In one, punters anchor on the visible rise; in the other, the edge comes from a winner kept at its peak. If raised runner-up quick returners also beat their price, the rise itself is what is mispriced. Columns: rating_change, last_position, days_since_last_run, is_handicap, last_is_handicap, sp_decimal.
Thread: H34: won easily last timetested by Test-mechanism: handicap runners-up raised 6 lb or more and back within 14 days beat their pricestoryraised 16 Oct 2026
Why do big-yard Irish top-two runners swing between 1.38 (earlier finding) and 0.64 (quality definition) depending on how 'big yard' is defined?
open0.45An Irish effect that flips with the yard definition is unstable and should not be used for selection. A per-trainer table on trainer_runs, trainer win-rate columns and course over the pooled Irish rows would show whether one or two yards drive it.
Thread: H199: favourite at a big price in a big handicaprace-bettingraised 16 Oct 2026
Does the live Snowden novice non-handicap system keep A/E above 1.05 at BSP in its forward record from 2026-10-13, judged as one pooled rule rather than by sub-slice?
open0.45Every child in the October 2025+ window (16, 27, 35 bets) is untestable, and the bumper child was killed only for thin periods and band concentration. The only measurement that can move this thread is the system's own forward record. It should be read at BSP because the edge disappears at SP. Columns: trainer, is_novice, is_handicap, race_type, sp_decimal (and bsp after the race).
Thread: H17: jamie snowden novicesnamed-trainerraised 16 Oct 2026
Do Snowden's bumper runners (the same young stock before they jump) also beat their prices?
testing0.45If the edge is about how the yard develops young horses, it should show in bumpers as well. If it does not, the edge is more about jumping or placing novices. Columns: trainer, race_type='Bumper', prior_runs, age.
Thread: H17: jamie snowden novicestested by Generalise: Snowden bumper runners beat their pricesnamed-trainerraised 16 Oct 2026
Did 5 of 5 survived attacks fail to predict a vault contradiction because the attacks all use the discovery window and so cannot detect in-window overfitting?
open0.45This is a method question. If attacks survived and the vault still contradicted, the attack suite may need a time-holdout attack. That would change how much weight the lab gives a FROZEN status across all threads. Columns: race date or season, alongside the rule's own predicates.
Thread: H126: lightly raced four year old returnsseasonalraised 16 Oct 2026
Is the long-gap turf-to-AW lift carried by a handful of 16/1+ winners rather than spread across the price bands?
open0.45The 16/1+ band scored 1.496 on only 11 wins. Outside March-April, 16/1+ scored 1.68 on 8 wins while 8/1 to 14/1 scored 0.59. Earlier in this thread lifts sat in the short bands. Columns: sp_decimal, sp_rank, surface, last_surface, days_since_last_run.
Thread: H126: lightly raced four year old returnsseasonalraised 16 Oct 2026
Is the third-period fade (A/E 1.05 on 125) the market catching up, or just noise on a small sample?
open0.45The live system's forward value depends on whether the edge is decaying. Read forward bets on SYS-young-stayer-marathon-handicap-chase by period and by sp_decimal band.
Thread: H175: young horse in a marathon chaseage-staminaraised 16 Oct 2026
When the frozen stiff-sprint rules (H-2b483ac5, H-cbd60237) build up forward bets, does the A/E hold near 1.2 in GB, and does it stay near 1.0 or below in Ireland?
open0.45Two vault reads (1.17 on 415, 1.25 on 506) agree on a modest edge. Every split that could explain it (course, draw, Ireland, size of drop) is too thin in discovery: 15, 20, 4 and 2 bets. Only forward data can grow these samples. A descriptive scan should read region, course, track_stiff_finish, distance_yards, last_distance_yards, last_win_distance_yards and sp_decimal on forward runners. Priority is moderate: the answer decides whether the system goes GB-only, but it needs time, not a new rule.
Why does Ireland keep showing no lift (IRE 0.74 on 25, earlier 0.85 on 23)? Is track_stiff_finish coding Irish courses whose rise comes before the sprint straight, or that have no straight sprint course at all?
testing0.45Ireland has now come out low three times. If the IRE flag marks a different kind of finish, the mechanism predicts no lift there, and the live system should be GB only on the mechanism's own terms rather than by slicing. Read track_stiff_finish, track_finish, track_straight_furlongs, track_undulations and course for IRE against GB.
Thread: H181: proven over further at a stiff tracktested by Rival: the Irish null is the stiff flag, not Ireland; the lift needs a short uphill straighttrack-staminaraised 16 Oct 2026
Is the parent's stiff-track lift in jumps racing different from the trip-drop lift at non-stiff jumps tracks?
open0.45The parent's 7,574 bets must be mostly Hurdle and Chase runners, because only about 412 are Flat races up to a mile. Comparing jumps at stiff and non-stiff tracks would show whether the hill matters anywhere outside sprints. Columns: race_type, track_stiff_finish, last_win_yards, distance_yards, track_finish.
Thread: H181: proven over further at a stiff tracktrack-staminaraised 16 Oct 2026
Is the lift a property of the handicap mark, or does it also show in Flat non-handicaps, where there is no mark to lag?
testing0.45The raised-mark cut (O-f62c7b90) points to the market doubting continued progress, not to the mark. If that is right, lightly raced four-year-old returners in conditions and maiden races should also beat their prices. Columns: is_handicap, is_maiden, age, prior_runs, days_since_last_run, race_class.
Thread: H126: lightly raced four year old returnstested by Lightly raced four-year-old returners in Flat non-handicaps: progress the market doubts, with no mark involvedseasonalraised 16 Oct 2026
Why was the middle period flat (1.01 on 1329 bets) when the outer periods were 1.21 and 1.20? And is the edge real only at Betfair SP (BSP ROI +1.6%, SP ROI -19.6%)?
open0.45An unstable middle period plus a profit that disappears at SP means the published system needs a check of its stability and of the price it can be bet at. A test splits by period, then by region, sp_rank and book_pct.
Thread: H126: lightly raced four year old returnsseasonalraised 16 Oct 2026
Does pooling all quick-returning last-time winners at 9/1+ across handicaps and non-handicaps give a consistent deficit in both periods and both countries, or does the non-handicap part only add noise?
open0.40The non-handicap replicate (106 bets, 2 wins) is too thin on its own, and its price-band pattern is the reverse of the handicap pool's. A split by is_handicap, region, race_type and period over last_position, days_since_last_run and sp_decimal would show whether one rule covers both race types.
Thread: Raised handicap winners sent off at 9/1+ are over-bet, at any gap since the winnegativeraised 16 Oct 2026
Do quick-returning last-time winners at 9/1+ fall short in non-handicaps too, where the winner tag is equally visible but no mark rise applies?
testing0.40A replication outside handicaps tells us whether the tag-plus-quick-return deficit is general, or tied to handicap penalties and reassessment. Columns: is_handicap, last_position, days_since_last_run, sp_decimal, race_type.
Thread: Raised handicap winners sent off at 9/1+ are over-bet, at any gap since the wintested by Replicate: quick-returning last-time winners at 9/1+ in non-handicaps are over-bettested by Generalise: quick-returning last-time winners at 9/1+ in any race are over-betnegativeraised 16 Oct 2026
Is the deficit stronger after bigger rises (rating_change >= 3) and does it hold beyond 14 days, as the mark mechanism requires?
open0.40Both refinements were planned in the spec but not reported. The mark mechanism predicts a dose-response with rise size and no dependence on the gap. Columns: rating_change, days_since_last_run, sp_decimal.
Thread: Raised handicap winners sent off at 9/1+ are over-bet, at any gap since the winnegativeraised 16 Oct 2026
Why does the lift sit in big-yard second favourites (pool 1.33) and not in big-yard favourites (1.06 on 54) or in the yard's own market choice among several entries (1.05 on 110)?
open0.40This pattern contradicts the split-attention mechanism and points to something about the second slot in the betting. It needs a by-rank breakdown of the big-yard pool. Columns: sp_rank, is_favourite, n_stable, stable_rank, trainer_runs.
Thread: H199: favourite at a big price in a big handicaprace-bettingraised 16 Oct 2026
Is Ireland's par in big-handicap short-end runners tied to how Irish books are made, for example a higher overround (book_pct) that flattens the front of the market?
open0.40Field size and yard size have both been ruled out, but the country gap remains. A descriptive scan of book_pct and favourite_sp by country for these runners would show whether the Irish front of the market is priced differently. Columns: book_pct, favourite_sp, course, field_size, sp_rank.
Thread: H199: favourite at a big price in a big handicaprace-bettingraised 16 Oct 2026
Is the gap between SP and BSP returns (-9.6% vs +16.6%) wider for this selection than for other favourites at the same price?
open0.40This decides whether the system can be played at SP or only on the exchange. The columns are book_pct, sp_decimal and field_size; BSP itself is post-race and can only be used to measure, never to select.
Thread: H199: favourite at a big price in a big handicaprace-bettingraised 16 Oct 2026
Does the young marathon-chase lift hold in fields under the big-field cut-off on a sample of 200+ bets, or does it sit mainly in large festival handicaps?
open0.40The niche reading outside big fields was 1.28 on 58 against a control of 0.85. That is suggestive but too thin. The answer decides whether Q-56f7bf65 can close. Columns: field_size, book_pct, course, race_class, sp_decimal.
Thread: H175: young horse in a marathon chaseage-staminaraised 16 Oct 2026
Is the up-to-3/1 lift in jumps departure rules just the general short-price end of the jumps market in this window (October 2025 on), shared by runners that completed last time?
testing0.40Three killed departure rules (fallers, going-well, weak-form going-well) all relied on their shortest price band, while the mid-price band scored 0.82. If all jumps runners at up to 3/1 score about the same A/E, the band concentration is favourite-longshot structure and the departure branch at normal prices can be closed for good. Columns: sp_decimal, race_type, last_fell_or_unseated, last_non_finish_code, last_position.
Thread: H98: brought down or unseatedtested by Rival: short-priced jumps runners that completed last time beat their price as much as short-priced departureslettersraised 16 Oct 2026
Is last_fell_going_well flagged at similar rates in GB and Ireland among fallers and unseaters, or does a difference in comment wording or coverage explain GB 1.27 against IRE 0.80?
open0.40The concentration comes from 42 winners and the mechanism gives no reason for a country split. Before reading it as a real effect, the descriptive scanner should compare the flag rate per fall or unseat by region, using last_fell_going_well, last_fell_or_unseated, region and last_phrases. A big gap in flag rates would point to a data artefact.
Thread: H98: brought down or unseatedlettersraised 16 Oct 2026
Is the returner lift specific to horses whose last win came in a handicap (a real mark rise), rather than to maiden or novice winners making a handicap debut?
open0.40The last_is_handicap = 1 partition scored 1.50 on 84, against 1.27 for the whole won-last-time group. This is the case the mark-rise mechanism names, but 84 bets cannot be read alone. A test would pool ages 3 to 5 and read last_is_handicap, last_position, last_official_rating, days_since_last_run and prior_runs.
Thread: H126: lightly raced four year old returnsseasonalraised 16 Oct 2026
Why did the broadened last-run rule shrink from about 1.54 in discovery to 1.17 in the vault while the capped last-win rule held 1.64? Is a last win over further the better stamina proof, or is this ordinary shrinkage?
open0.40The two rules overlap heavily. If last-win runners carry the vault lift and the last-run-only runners are near 1.0, our earlier finding that stamina, not class, drives it is wrong. Columns: last_win_yards, last_distance_yards, last_beaten_lengths, sp_decimal, track_stiff_finish.
Thread: H181: proven over further at a stiff tracktrack-staminaraised 16 Oct 2026
Do `last_code IN ('BD','UR')` and `last_non_finish_code IN ('BD','UR')` select the same runners?
open0.40The hypothesis was measured on `last_code`, but the system is published on `last_non_finish_code`. If the two columns differ (for example, if `last_code` holds the jumping-shock Chase/Hurdle code), then the live record measures a different population from the one that was graded GREEN.
Thread: H98: brought down or unseatedlettersraised 16 Oct 2026
Does A/E fall steadily as prior_runs rises among four-year-old spring returners (0-5, 6-10, 11-15, 16+), or is there a step at about 10 runs?
open0.40Exposed returners scored 0.80 on 191 against the base's 1.13, but the fewest-runs cut failed on price-band concentration. A descriptive scan of A/E by prior_runs band would show whether light racing works as a gradient (supporting an unexposed-improver story) or whether the 10-run cut-off is arbitrary. Columns: prior_runs, flat_runs, age, days_since_last_run, month, is_handicap, sp_decimal.
Thread: H126: lightly raced four year old returnsseasonalraised 16 Oct 2026
Is the effect stronger at age 4 than at age 5, as the mechanism predicts, since sending a four-year-old over fences is the rarer and more deliberate choice?
open0.40The mechanism says rarity signals intent, so the effect should be strongest at age = 4. Read age, trainer and region; four-year-old chasers are often French-bred and may cluster in a few yards, which would need checking.
Thread: H45: young horses in novice chasesageraised 16 Oct 2026
Why would poor jumping be over-bet in last-time winners but not in placed horses: is it the extra market support a last-time winner gets?
open0.35Poor jumpers that won scored 0.78; those placed first to third scored 1.013. If the mechanism is market overreaction to the win, poor-jumping winners should be shorter today than last time compared with good-jumping winners. Read sp_decimal, last_sp_decimal, last_position and last_jump_index_pct.
Thread: Chase-bound winners that jumped poorly when winning are over-betnegativeraised 16 Oct 2026
Is the long-price winner deficit concentrated where the market has turned against the horse: short last time (last_sp_decimal <= 5) and now 9/1+?
open0.35If the price drift reflects information the stable or the market has (going, trip, a hidden problem), drifting winners should be over-bet more than winners that were long last time too. Being long last time is a visible trait. A big drift may carry information the residual winner-tag money ignores. Columns: last_sp_decimal, sp_decimal, last_position, days_since_last_run.
Thread: Raised handicap winners sent off at 9/1+ are over-bet, at any gap since the winnegativeraised 16 Oct 2026
Is a low top-speed percentile last time a generic mild negative for every runner, so that a 'slow winner' has no specific effect beyond it?
open0.35The slow-winner pool scored 0.961. Close-up losers with a slow top speed (IQ150d) scored 0.967 on 9,518. If the two are the same, the winner angle is dead. A descriptive scan could compare A/E by bands of last_top_speed_pct (or l_top_speed_pct / l_top_speed_rel), split by last_position = 1 versus > 1 and by sp_rank <= 3.
Thread: Last-time winners that were slow in the race are over-bet at the front of the marketnegativeraised 16 Oct 2026
Do big-rise quick returners at 16/1 or bigger share a visible obstacle (a class step-up, a trip change, a code switch) that explains why they fail?
testing0.35Long-priced raised quick winners scored 0.57 on 255, with GB at 0.46. If the failures share a visible step-up, a price cap stands in for a real cause; if they do not, it is a dismissal the market gets right. Columns: race_class, last_race_class, distance_yards, last_distance_yards, race_type, last_race_type, sp_decimal, region.
Thread: H34: won easily last timetested by Big-rise quick-returning handicap winners stepping up 15%+ in trip are over-betstoryraised 16 Oct 2026
Do the non-handicap H19 runners (about 28 bets) carry an outsized share of the parent's lift? If so, the market, not the handicapper, is what misreads the run.
open0.35The parent scored 1.28 on 218 and the handicap subset 1.145 on 190, which implies a very high A/E on the small non-handicap residue. That would shift the mechanism from the mark to the market. The sample is thin, so it is a monitoring question for now. Columns: is_handicap, is_maiden, is_novice, last_is_handicap, last_momentum_interrupted, last_beaten_lengths.
Thread: H19: momentum interrupted beaten narrowlystoryraised 16 Oct 2026
Is the 1.20 vault reading for young, lightly raced marathon chasers spread across trainers and races, or does it come from a few yards that target these races?
open0.35A supportive vault result on 290 bets with p 0.136 could rest on a handful of yards that are good at placing young stayers (e.g. a few GB trainers). GB carried discovery (1.31 on 56, IRE 0.78 on 19). Columns: trainer, course, region, prior_chase_runs, field_size.
Thread: H175: young horse in a marathon chaseage-staminaraised 16 Oct 2026
Is the young marathon-chase lift a GB-only effect, given IRE scored 0.78 on 19 against GB 1.31 on 56 in the unexposed slice?
open0.35If IRE runs below par across the parent system too, the system should be scoped to GB, or the mechanism (for example, how GB and IRE trainers campaign young stayers) needs revising. A test would read region, age, chase_runs, distance_yards and trainer on the parent population.
Thread: H175: young horse in a marathon chaseage-staminaraised 16 Oct 2026
Are older (8+) runners more than 7 lb above bottom weight in 3m4f+ handicap chases mildly over-bet, the mirror of the young lift, given their 0.90 on 1129 in the vault?
testing0.35The vault result was the opposite of the prediction (rule 10). Older class horses may take money as proven stayers. It cannot be tested again on vault years, so it needs forward or fresh data. Columns: age, weight_lbs, bottom_weight_lbs, sp_decimal, distance_yards, is_handicap.
Thread: H175: young horse in a marathon chasetested by Young horses in 3m+ handicap hurdles beat their price as marathon chasers doage-staminaraised 16 Oct 2026
Does the 1.08 for non-class-dropping turf-to-AW switchers survive in data outside the discovery window, or is it the same artefact as H-98ae1db9?
open0.35The surprise rests on 3797 bets with a CI that includes 1.0, drawn from the window the vault has just contradicted. A descriptive look at a later window (columns: last_surface, surface, race_class, last_race_class, region) would settle whether there is anything to test before spending a vault query.
Thread: H126: lightly raced four year old returnsseasonalraised 16 Oct 2026
Are near-bottom-weight runners in big marathon handicaps often out of the handicap, so they are carrying more than their rating deserves?
open0.35In Nationals-type races the weights are compressed. Horses on the minimum weight can be running off a mark below their real one, which the racecard weight hides. A test would read official_rating, weight_lbs, top_weight_lbs, bottom_weight_lbs and field_size. official_rating is only available from 2025-10 onward, so the sample is limited.
Thread: H175: young horse in a marathon chaseage-staminaraised 16 Oct 2026
Does the turf-to-AW lift exist in Ireland at all, or is it a GB AW-programme effect?
open0.35Three AW rules in this thread underperform in Ireland: IRE 0.449 on 40 (kill reason), short-gap IRE 0.87 on 473, and long-gap IRE 1.128 on only 93. If the effect is GB-only, the lab should restrict the population up front instead of re-killing rules on Ireland. Columns: region, course, surface, last_surface.
Thread: H126: lightly raced four year old returnsseasonalraised 16 Oct 2026
Are horses at or near bottom weight in staying handicap chases (3m+) over-bet more generally, for example because they run from out of the handicap or are declining horses with falling marks?
open0.35The 0.48 on 124 is a single slice (p 0.08 on the opposite side). Only a wider population can tell whether it is noise or a real over-bet on bottom-weights. Columns: weight_lbs, bottom_weight_lbs, last_official_rating, last_weight_lbs, distance_yards, age, sp_decimal.
Thread: H175: young horse in a marathon chaseage-staminaraised 16 Oct 2026
Is the repeated up-to-7/1 concentration in this thread (step-up rule 1.36; small-field rule falsified on the same band) a real property of the young-stayer edge, or the usual concentration noise of small samples?
open0.35Two rules that share runners both put their lift in sp_decimal <= 8.0, with the rest at 0.86 to 0.89. The rules overlap, so this is one pattern, not two confirmations. The answer should come from the live system's forward bets read by sp_decimal band, not from slicing a new rule.
Thread: H175: young horse in a marathon chaseage-staminaraised 16 Oct 2026
Within the non-four-year-old second-tier lift, is it the three-year-olds or the exposed older horses (prior_runs > 10) that beat their price?
open0.35The rule had no prior_runs cap, while lightly raced age 5+ returners scored 0.64 on 119. If exposed older horses carry the lift, the idea of lightly raced horses improving behind their mark is irrelevant, and the effect looks more like fitness or the stable's intent after a break. Columns: age, prior_runs, sp_decimal, days_since_last_run.
Thread: H126: lightly raced four year old returnsseasonalraised 16 Oct 2026
Is the fade after 27 April in the fancied-returner rule (1.18 to 0.97) a May effect that repeats each spring, or just noise in one season?
open0.35The second period is almost all May 2026. That lines up with the March-April sharpen (1.29) against May (1.06), but the 4/1-cap neighbour held 1.20 after 27 April. A month-by-month read of month, sp_decimal and days_since_last_run would show whether timing or price band drives the split.
Thread: H126: lightly raced four year old returnsseasonalraised 16 Oct 2026
Is the 1.07 lift on 971 lightly raced three- and four-year-old non-winner returners a stable feature, or a property of the 2026 spring alone?
open0.35The non-winner control matched the base H126 level and was flat across both periods (1.07, 1.07), with the best band at up to 14/1 (1.33 on 290). Only one spring is in discovery, so the vault and the forward period must decide. Columns: last_position, age, prior_runs, days_since_last_run, sp_decimal.
Thread: H126: lightly raced four year old returnsseasonalraised 16 Oct 2026
Are country kills on 34 to 37 Irish bets informative, given that Irish slices range from 0.91 to 1.90 across nested wide-trip rules?
open0.35H-77a1708f was killed on IRE 0.91 (36 bets), and the all-weather-today test on IRE 0.90 (37). Yet the drift rule shows IRE at 1.90 on 34. A method check on how spread out region-level A/E is at n of about 35 (read region, sp_decimal) would show whether these kills discard real edges.
Thread: IQ140a: wide trip with figuresoverlayraised 16 Oct 2026
Do short-priced lightly raced four-year-old returners in non-handicaps (up to 3/1) beat their prices because the yard's confidence shows in the market, or is that band one of many segments?
testing0.35O-d5ec2505 killed the non-handicap version because only the up-to-3/1 band carried it (rest 1.00). If short-priced returners also do best in handicaps, the shared cause is market-signalled readiness, not the mark. A test reads sp_decimal, sp_rank, is_handicap, race_type, days_since_last_run, prior_runs and age.
Thread: H126: lightly raced four year old returnstested by Short-priced first-time-out four-year-olds: rival to the handicap-mark storyseasonalraised 16 Oct 2026
Is the returner lift concentrated in a few big yards whose four-year-olds return ready, rather than spread across trainers?
open0.35Discovery samples are small (140 all-weather bets, 41 raised-mark bets), so a handful of yards could carry them. A trainer concentration scan on the H126 base would rule a yard effect in or out. Columns: trainer, prior_runs, days_since_last_run, sp_rank.
Thread: H126: lightly raced four year old returnsseasonalraised 16 Oct 2026
Within the full parent base on the Flat, does A/E differ between runners drawn in the low half and the high half today?
open0.35This is the draw half of Q-b8b1b52c, which no test has yet measured like-for-like. The earlier low-draw subset (1.12 on 456) dropped the parent's exclusion clause and was carried by 16/1+. A test would read draw, field_size and race_type.
Thread: IQ140a: wide trip with figuresoverlayraised 16 Oct 2026
Over hurdles, are horses that won despite a poor Jump Index underrated, with the market marking down the untidy jumping more than it should when the horse clearly has ability?
open0.30The hurdle generalisation came out the other way: 1.074 on 396, steady across periods and countries. The CI (0.86 to 1.33) is wide, and 3 wins at 16/1+ carry much of the gain, so the price spread needs checking first. Columns: last_jump_index_pct, last_position, sp_rank, sp_decimal, last_sp_decimal, hurdle_runs, jumping_errors_last3. Priority is moderate because a positive edge would need about 1.10 on many more bets.
Thread: Chase-bound winners that jumped poorly when winning are over-betnegativeraised 16 Oct 2026
Is the location of the deficit by price band stable at all, given that it sat at 9/1–14/1 for raised winners but at 16/1+ for unraised ones?
open0.30The two pools show opposite price-band patterns, so the earlier '9/1–14/1, not favourite-longshot' finding may be noise. A descriptive scan of sp_decimal bands by race_type and region across all last-time handicap winners would show this before any band-restricted rule is written. Columns: sp_decimal, race_type, region, rating_change.
Thread: Raised handicap winners sent off at 9/1+ are over-bet, at any gap since the winnegativeraised 16 Oct 2026
Is the frozen within-35-days version of H19 (1.19 on 200) building any forward record, or is it waiting on turf-season volume?
open0.30Every sub-slice of H19 has died from thin samples or one weak period. The forward record of the shadow and live systems is now the only clean judge. A scanner should count forward bets by month and by surface before the lab spends more slices. Columns: days_since_last_run, surface, last_momentum_interrupted.
Thread: H19: momentum interrupted beaten narrowlystoryraised 16 Oct 2026
As more data builds up, does the checked-and-recovered subgroup of H19 keep beating the rest of H19?
open0.301.59 on 31 (CI 0.73 to 3.02) fits the hidden-ability mechanism but cannot yet tell recovered from unrecovered. It needs about 100 bets. Columns: last_recovered, last_strong_finish, last_momentum_interrupted, last_beaten_lengths.
Thread: H19: momentum interrupted beaten narrowlystoryraised 16 Oct 2026
Are checked runners beaten 2 to 5 lengths over-bet when the market shortens them on the excuse, i.e. is visible trouble over-credited once the margin is wider?
open0.30The 2-to-5-length group came out at 0.80 against a 1.03 control, the opposite of the prediction. If punters credit the excuse, the shortened subset should be worst. Much of the shortfall is 0 winners on 61 bets at 16/1 or bigger, so this is a lead, not a finding. Columns: sp_decimal, last_sp_decimal, last_beaten_lengths, last_momentum_interrupted.
Thread: H19: momentum interrupted beaten narrowlystoryraised 16 Oct 2026
Among young, lightly raced marathon chasers, are those whose recent runs show jumping errors the ones that beat their price, as the jumping-improvement mechanism says?
open0.30The mechanism says the short chase record is made to look worse by early jumping errors that the horse then grows out of. If so, the lift should sit with runners whose last runs carried errors or poor jump figures, not with clean jumpers. The sample is thin (about 75 bets in discovery), so a descriptive scan should look first. Columns: jumping_errors_last3, last_jumping_error, last_fell_or_unseated, l_jump_index_pct, last_jump_index_pct, prior_chase_runs.
Thread: H175: young horse in a marathon chaseage-staminaraised 16 Oct 2026
Does the 4/1 to 7/1 concentration recur in every marathon handicap slice because a few well-backed staying types win at those prices, or is it noise across small slices?
open0.30Three separate slices put their lift in the up to 7/1 band (1.33, 1.55 and the young step-ups at 1.36), with up to 3/1 and 8/1 to 14/1 near or below par. If this pattern is a structural feature of marathon pricing, slice results should be read with it in mind. Columns: sp_decimal, sp_rank, field_size, distance_yards.
Thread: H175: young horse in a marathon chaseage-staminaraised 16 Oct 2026
Do Irish all-weather (Dundalk) returners really fail, or is 0.45 on 40 just noise in a thread that is GB-led everywhere?
open0.30Ireland weakness has ended several hypotheses in this thread. Dundalk returners year-round (128 bets, 0.82) give the only larger sample. Columns: region, course, surface, days_since_last_run.
Thread: H126: lightly raced four year old returnsseasonalraised 16 Oct 2026
Is the Irish lift for jumps departures (fallers 1.21, not-going-well departures 1.27) a real Irish effect, or noise that flips between sibling rules (going-well IRE 0.80, weak-form IRE 0.96)?
open0.30The country concentration points different ways in rules that share most of their runners. One descriptive check would settle it: A/E for all Irish jumps departures (last_fell_or_unseated = 1) by sp_decimal band and by period. If it does not hold across all departures and both periods, stop slicing by region. Columns: region, last_fell_or_unseated, last_non_finish_code, sp_decimal, field_size, is_handicap.
Thread: H98: brought down or unseatedlettersraised 16 Oct 2026
Is the repeated Irish null for spring returners structural (big yards' returners in Ireland priced sharply) or just small samples of 50 to 80 bets?
open0.30Ireland scored 0.88 on 80 here and 0.887 on 50 in the sp_rank rule, while GB carried 1.12. A descriptive scan of returners by region and trainer volume (region, trainer, sp_decimal, days_since_last_run, age) would show whether the Irish shortfall sits in a few big yards or is spread evenly.
Thread: H126: lightly raced four year old returnsseasonalraised 16 Oct 2026
Why do the going-well rules repeatedly split GB high and Ireland low (1.26 against 0.80, and 1.56 against 0.96)? Is last_fell_going_well coded differently for Irish comments?
open0.30The mechanism predicts no country split. A difference in flag coverage would mean the Irish rows are mislabelled rather than unpriced. Columns: region, last_fell_going_well, last_fell_or_unseated, last_phrases.
Thread: H98: brought down or unseatedlettersraised 16 Oct 2026
Is the early-spring lift really confined to returners at 14/1 or shorter, given that 16/1+ went 0.79 on 114 in March–April but 1.19 on 245 in the earlier unfancied cut?
open0.30The longshot band contradicts itself across cuts in this thread, and the vault already showed that the 14/1 cap does not sharpen H126. This should be read by the descriptive scanner on sp_decimal and sp_rank by month before anyone proposes a rule, so that a near miss is not sliced until it passes.
Thread: H126: lightly raced four year old returnsseasonalraised 16 Oct 2026
The vault had 141 price-capped bets against only 31 in discovery: is the stiff-sprint population growing over time, or did the windows differ in the share of stiff-track sprint meetings?
open0.30The falsification attacks could not compare periods (too thin), so the forward record will decide. Knowing whether the bet rate is stable tells us how fast the live system can be judged. Columns: course, race_type, distance_yards, track_stiff_finish, region.
Thread: H181: proven over further at a stiff tracktrack-staminaraised 16 Oct 2026
Are Irish splits of under 100 bets in this thread informative at all, given they flip from 1.48 to 0.73 across overlapping cuts?
open0.30H-3c53e905 was killed on Ireland at 0.887 on 50 bets. The sibling capped rule scored Ireland 1.48 on 72 (1.60 on 50 when tightened to 11/1). A test would compare, by region, the Irish A/E of the H126 base against its top three in the betting (sp_rank) and its 14/1-or-shorter subset (sp_decimal), with bootstrap intervals. This is a method question for the falsification gate more than for the edge.
Thread: H126: lightly raced four year old returnsseasonalraised 16 Oct 2026
With more data, do lightly raced four-year-old returners in Flat non-handicaps beat their prices at 16/1 or shorter, or is the 1.07 just two longshot winners?
open0.30The non-handicap niche (O-c3a86a60) is the only test of whether the effect exists without a mark. Its BSP ROI rests on 2 winners from 64 at 16/1 or bigger. This needs forward data, not slicing now. Columns: is_handicap, is_maiden, is_novice, race_class, sp_decimal, region.
Thread: H126: lightly raced four year old returnsseasonalraised 16 Oct 2026
Does the lift sit mainly in the market's first four or five in the betting, with returners at 16/1 or bigger simply unready?
open0.30The all-weather version had 0 wins from 48 bets at 16/1 or bigger (O-a51c66ea), and the fewest-runs cut lived in one price band. This repeated pattern should be described before anyone slices on price. Columns: sp_rank, sp_decimal, days_since_last_run.
Thread: H126: lightly raced four year old returnsseasonalraised 16 Oct 2026
Are the 16/1+ winners that lift the big-field and sharp-track subsets (6 wins each, A/E about 2.0) largely the same horses?
open0.30If the two subsets share their few longshot winners, they are not independent support. The lab should then judge on the at-or-under 14/1 A/E only. A test would read sp_decimal, last_field_size and last_track_sharpness.
Thread: IQ140a: wide trip with figuresoverlayraised 16 Oct 2026
Is the poor-jumper deficit over fences an Irish chase effect, not a jumping effect? Do Irish chase runners with a poor last-run Jump Index underperform whatever their last finishing position, while GB runners do not?
open0.25Both pools below 1 rest on Ireland (IRE 0.45 in the parent, 0.44 on 57 in the hurdle-to-fence sharpen), while GB sat at 0.86 to 1.03. A descriptive scan should read region, race_type, last_jump_index_pct, last_position and chase_runs before anyone forms a hypothesis. A region split found after the fact is a lead, not a finding, and the Irish samples are small.
Thread: Chase-bound winners that jumped poorly when winning are over-betnegativeraised 16 Oct 2026
Is the 0 winners from 17 at 16/1 or bigger in the raised easy-winner rule a structural cut, meaning the market correctly marks those runners as a class or trip step-up, or is it noise?
testing0.25With p of about 0.05 per bet, 0 from 17 is weak evidence on its own. But the parent shadow also had 0 from 24 at 16/1+, so the pattern repeats. A price cap would only be justified if the long-priced runners share a visible handicap (class rise, trip change). Columns: sp_decimal, race_class, last_race_class, distance_yards, last_distance_yards.
Thread: H34: won easily last timetested by Long-priced raised handicap winners turned out within 14 days are over-betstoryraised 16 Oct 2026
Is the older higher-weighted 16/1+ lift (1.46 on 110 from 4 wins) anything more than a few big-field shock winners?
open0.25The lift rests on 4 winners, probably in big festival fields. Before treating it as signal, check how it splits by field_size, course and book_pct. A test would read field_size >= 16, book_pct, course, sp_decimal.
Thread: H175: young horse in a marathon chaseage-staminaraised 16 Oct 2026
Is the going-well lift concentrated in the roughly 2/1 to 7/1 bands, given that the 'up to 14/1' band scored 0.91 on 98 and evens-or-shorter 0.56 on 10?
open0.25Treat this as a lead from one breakdown, not something to slice. It matters for reading the forward record of any going-well rule. A descriptive pass over sp_decimal and sp_rank bands within last_fell_going_well = 1 would show whether the shape is smooth or comes from a few winners.
Thread: H98: brought down or unseatedlettersraised 16 Oct 2026
Is the Irish shortfall specific to Dundalk all-weather, or are Irish wide runners weak across the board once turf is included?
open0.25The kill rests on 37 Irish all-weather bets (5 wins, A/E 0.90) and 30 all-weather to all-weather bets (0.75). On turf the same base scored 1.59 on 138 Irish bets. A descriptive split by region x surface (region, surface, course) would show whether the live system needs a GB-or-turf condition or whether this is noise. Treat it as a scan, not a refinement to pass.
Thread: IQ140a: wide trip with figuresoverlayraised 16 Oct 2026
Is the Irish part of H126 weaker for unraised horses because the Irish handicapper treats light three-year-old form differently?
open0.25Unraised Irish returners won 0 of 42, while Irish turf returners overall scored 1.47 on 121. Region interacts with rating_change. A test reads region, rating_change and surface. The samples are small, so information gain is limited.
Thread: H126: lightly raced four year old returnsseasonalraised 16 Oct 2026
Is the 0-from-107 at 16/1+ among young lightly raced 3m to 3m3f chasers different from the general strike rate of 16/1+ chasers at those trips?
open0.20O-99a444ce: the whole negative lean may come from longshots, and longshot weakness is a trait the market already prices. If 16/1+ runners of all ages at 3m to 3m3f do as badly, this is plain longshot bias and the thread should drop the sub-3m4f arm. A test would read sp_decimal, age, chase_runs, distance_yards and race_type. The scanner can check this cheaply, but it is unlikely to change what the lab does.
Thread: H175: young horse in a marathon chaseage-staminaraised 16 Oct 2026
Is the late-April window (28-30 April, 1.41 on 101) a real step before May (0.986 on 792), or noise from a few wins?
open0.20The thread's fade story has moved twice: first to 27 April, now to the month boundary. Before anyone builds rules on dates, the lab should know whether returner A/E changes smoothly by day of season or only in one window of a single spring. Columns: month, days_since_last_run, race date, course.
Thread: H126: lightly raced four year old returnsseasonalraised 16 Oct 2026
Does the stiff-sprint edge need a past win over further, or does any recent run over further carry the same signal?
testingraised with hypothesis 'Generalise: any proven stamina (last run or last win 330+ yards further) in stif'
Thread: H181: proven over further at a stiff tracktested by Generalise: any proven stamina (last run or last win 330+ yards further) in stiff Flat sprints at 14/1 or shortertrack-staminaraised 16 Oct 2026
Thread: H181: proven over further at a stiff tracktrack-staminaraised 16 Oct 2026